mindee/src/parsing/common/document.js

Mindee Client Library for Node.js

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.Document = void 0;
const extras_1 = require("./extras");
const extras_2 = require("./extras/extras");
const ocr_1 = require("./ocr");
const ragExtra_1 = require("./extras/ragExtra");
/**
 * Document prediction wrapper class. Holds the results of a parsed document.
 * @typeParam T an extension of an `Inference`. Mandatory in order to properly create an inference.
 */
class Document {
    /**
     *
     * @param inferenceClass constructor signature for an inference.
     * @param httpResponse raw http response.
     */
    constructor(inferenceClass, httpResponse) {
        this.id = httpResponse?.id ?? "";
        this.filename = httpResponse?.name ?? "";
        this.ocr = httpResponse.ocr && Object.keys(httpResponse.ocr).length > 0 ? new ocr_1.Ocr(httpResponse.ocr) : undefined;
        this.inference = new inferenceClass(httpResponse["inference"]);
        if (httpResponse["extras"] &&
            Object.keys(httpResponse["extras"].length > 0)) {
            const extras = {};
            Object.entries(httpResponse["extras"]).forEach(([extraKey, extraValue]) => {
                switch (extraKey) {
                    case "cropper":
                        extras["cropper"] = new extras_1.CropperExtra(extraValue);
                        break;
                    case "full_text_ocr":
                        extras["fullTextOcr"] = new extras_1.FullTextOcrExtra(extraValue);
                        break;
                    case "rag":
                        extras["rag"] = new ragExtra_1.RAGExtra(extraValue);
                        break;
                }
            });
            this.extras = new extras_2.Extras(extras);
        }
        if (!this.extras || !("fullTextOcr" in this.extras) || this.extras["full_text_ocr"].toString().length === 0) {
            this.injectFullTextOcr(httpResponse);
        }
        this.nPages = httpResponse["n_pages"];
    }
    /**
     * Default string representation.
     */
    toString() {
        return `########\nDocument\n########
:Mindee ID: ${this.id}
:Filename: ${this.filename}

${this.inference?.toString()}`;
    }
    injectFullTextOcr(rawPrediction) {
        if (rawPrediction["inference"]["pages"].length < 1 ||
            !("extras" in rawPrediction["inference"]["pages"][0]) ||
            rawPrediction["inference"]["pages"][0]["extras"].length < 1 ||
            !("full_text_ocr" in rawPrediction["inference"]["pages"][0]["extras"]) ||
            !rawPrediction["inference"]["pages"][0]["extras"]["full_text_ocr"] ||
            !("content" in rawPrediction["inference"]["pages"][0]["extras"]["full_text_ocr"]) ||
            !rawPrediction["inference"]["pages"][0]["extras"]["full_text_ocr"]["content"]) {
            return;
        }
        const fullTextOcr = rawPrediction["inference"]["pages"].filter((e) => "extras" in e).map((e) => e["extras"]["full_text_ocr"]["content"]).join("\n");
        const artificialTextObj = { "content": fullTextOcr.length > 0 ? fullTextOcr : "" };
        if (!this.extras) {
            this.extras = new extras_2.Extras({ "fullTextOcr": new extras_1.FullTextOcrExtra(artificialTextObj) });
        }
        else {
            this.extras["fullTextOcr"] = new extras_1.FullTextOcrExtra(artificialTextObj);
        }
    }
}
exports.Document = Document;